What the new wave of agentic AI demands from CEOs

But with recent advances in artificial intelligence, this basic logic has begun to change. “For the first time, technology is now capable of this Doing the work“, Nvidia CEO Jensen Huang recently noted.[For example]Inside each robotaxi is an invisible AI driver. This driver does the work. The tool he uses is the car.”

This idea embodies the transformation taking place today. AI is no longer just a tool for human use: it has become a tool Active operator and coordinator “Action” itself, is capable not only of predicting and generating, but also of planning, implementing and learning. This emerging layer – “agent” AI – represents the next wave of AI. Agents can coordinate across courses of action, make decisions, and adapt with experience. In doing so, they also blur the line between machine and teammate.

For business leaders, this means that agentic AI turns basic management calculations about technology deployment on its head. Their mission is no longer just to install smarter tools, but to guide organizations where entire parts of the workforce are complex, distributed, and constantly evolving. With agents in place, companies must rethink their makeup: how to design work, how to make decisions, and how to create value when AI can execute on its own. How organizations redesign themselves around these powerful capabilities will determine whether AI becomes not just a more efficient technology, but a new basis for strategic differentiation altogether.

To better understand how executives are navigating this shift, BCG and MIT Sloan Management Review Conducted a global study of over 2,000 leaders from over 100 countries. The results show that while organizations are rapidly exploring agentic AI, most organizations still need to define the general strategies and operating models needed to integrate AI agents into their daily operations.

Organizational challenge: redesigning the organization

The perceived dual identity of Agentic AI – as machine and teammate – creates tensions that traditional management frameworks cannot easily resolve. Leaders cannot completely eliminate these tensions; They must instead learn how to manage it. Four organizational tensions stand out:

1. Scalability versus adaptability. Machines scale predictably, while people adapt dynamically. Agent AI can do both, requiring new organizational design principles capable of balancing efficiency and flexibility across workflows.
2. Experience versus expediency. Leaders must balance building long-term capabilities with moving quickly enough to seize short-term opportunities in a rapidly changing technology landscape.
3. Supervision versus autonomy. Agent AI requires oversight not only of outputs, but also of actions; Organizations must decide when humans stay informed and when agents act independently, with clear accountability structures for each.
4. Retrofitting versus reimagining. Leaders must choose when to layer AI over existing processes for immediate benefit and when to rebuild end-to-end workflows around the agent’s capabilities.

Advanced companies do not resolve these tensions directly. Instead, they embrace it, redesigning systems, governance, and roles to turn friction into forward momentum. They see the complexity of agent AI as a feature to be harnessed, not a bug to be fixed.

What should leaders do now?

For CEOs, the challenge now is to figure out how to lead an organization where technology works hand-in-hand with people. Managing this new class of systems requires different frameworks than previous waves of AI. While predictive AI has helped organizations analysis Help faster, better and generative AI creates Faster and better, agent AI now lets them do it It works Faster and better, through planning, implementation and improvement on their own. This transformation turns traditional management approaches upside down and requires new rules of leadership.

Reimagine work, not just workflow. In predictive or generative AI, the driving task is to incorporate models into the workflow. But agentic AI requires something different: it doesn’t just carry out a process; Dynamically reimagined. Because agents plan, act, and learn iteratively, they can discover new, and often better, ways to achieve the same goal.

Historically, many work processes were designed to have humans emulate machine-like precision and predictability: each step was standardized so that the work could be repeated reliably. However, effective systems reflect this logic: leaders need only specify the desired inputs and outcomes. The action that occurs between the start and end points is then organica living system that improves itself in real time.

But most organizations still treat AI as a layer on top of existing workflows, as a tool at its core. To harness the true potential of agentic AI, leaders must start by identifying a few high-value, end-to-end processes—where speed of decision-making, cross-functional coordination, and learning feedback loops are most important—and redesign them around how humans and agents learn and work together. The opportunity is to create systems that are able to scale predictably and adapt dynamically, not one or the other.

Guide actions, not just decisions. Previous waves of AI required oversight of the output; Agent AI requires supervision procedures. These systems can operate independently, but not all procedures carry the same risks. This makes the leadership challenge broader than simply defining decision-making rights. It defines how agents operate within an organization: what data they can see, what systems they can operate, and how and to what extent their choices ripple across the organization. While leaders will need to identify which categories of decisions will remain human-only, which can be delegated to agents, and which require cooperation between the two, the overall focus should be around setting boundaries for agents Behaviors.

Thus, governance can no longer remain a fixed policy; Must be flexible to context and risks. Just as leaders train people, they will also need to train agents—to determine what information they need, what goals to improve upon, and when to escalate uncertainty to human judgment. Companies that adopt these new approaches to governance will be able to build trust, both internally and with regulators, by making accountability transparent even when machines are capable of implementing it.

Rethink structures and talent. Generative AI has changed the way people work; Agent AI is changing how organizations are organized. When agents can coordinate action and information flow, the traditional middle layer devoted to supervision will shrink. This isn’t a replacement story, it’s a redesign. The next generation of leaders will be coordinators, not supervisors: people who can combine business judgment, technical fluency, and ethical awareness to guide mixed teams of humans and agents. Companies must start now by planning for broader hierarchies, less routine roles, and new career paths that reward coordination and innovation over task execution.

Institutionalizing learning for humans and agents. Like people, agents drift, learn, and most importantly, improve through feedback. Every action, interaction and correction makes them more capable. But this improvement depends on people staying involved, not to control every step, but to help systems learn faster and better.

To achieve this, leaders must create continuous learning loops that connect humans and agents. Employees must learn how to work with agents—how to improve them, critique them, and adapt to their evolving capabilities—while agents improve through those same interactions, through preparation, monitoring, retraining, and even “retirement.”

Organizations that approach this as a co-development process – where people shape how agents learn and evolve how people work – will see the greatest gains. Managing this loop requires viewing both humans and agents as learners, and creating structures for continuous training, retraining, and knowledge sharing. When this process is done right, the organization itself becomes a constantly improving system, one that becomes smarter every time humans interact with its agents.

Building radical resilience. Traditional turnaround programs are designed for predictability. However, the AI ​​agent moves too quickly for them to keep up. Leaders need organizations that can continually adapt, financially, operationally, and culturally. But adaptability in the age of agents is not just about keeping up with the faster technology cycle, it is also about being prepared to evolve as your organization learns alongside its agents. Each new capability can reshape responsibilities, decision flows, and even what “good performance” looks like.

Leaders will need to treat resilience not as crisis management, but as an organizing principle. This means budgeting for continuous reinvestment, building modular structures that allow for reconfiguration of functions as agents take on new roles, and cultivating cultures where experimentation is routine rather than exceptional. The Agentic AI program rewards organizations that can drive continuous, disruptive change. This kind of “agent centricity” means reassigning talent, modernizing processes, and updating governance in response to what the system itself is learning. The most agile companies will see adaptability not as a defensive response, but rather as a defining source of advantage.

Agent institution

For many years, the story of AI has been one of automation, that is, doing the same work faster, cheaper and with fewer people. But that era is coming to an end. Agent AI is changing the nature of value because it can reshape the organization itself: how it learns, collaborates, and evolves. The next frontier is radical redesign, not iteration.

The real opportunity is to create an organization that can continuously reinvent itself, where agentic AI becomes the connective tissue – connecting knowledge, decision-making and adaptation into a single living system. This is the basis of what we call it Agent enterprise operating system: A model in which human creativity and machine initiative evolve together, dynamically redesigning the way a company operates. Companies that embrace this transformation will outperform those still chasing efficiency, and will be the ones who define how value, capability, and competition operate in the age of AI.

He reads last luck Columns by François Candelon.

François CandelonHe is a partner at private equity firm Seven2 and former global director of the BCG Henderson Institute.